HISTORY

1.0.2: Jan 8, 2013
  - Fixed issue about detecting pairing when a read group mentioned in sam
    headers does not have enough mapped reads to infer distribution. Before,
    the read group wasn't mentioned in the output, possibly causing a crash.
    Now, such read groups will use a sensible default pairing information.
  - Improved pairing information aggregation. If the same read group appears
    in several sam input files, the one detected with the smallest standard
    deviation will be used.
  - Fixed typos in README.
  - Removed hard-coding of paths to awk and python. Before, we assumed they are
    is /usr/bin. Now, we find them through the environment (first in PATH).
  - Removed use of stderr inside a script, in favour of another variable file
    descriptor. This could have caused crashes if other things used stderr.
  - Changed mechanism to autodetect PHRED if not given.
  - Fixed read length detection when not enough reads. Before: -1, now: 100.
  - Fixed issue involving several read sets with the same read group when some
    mappings were lost due to a file naming conflict.

1.0.3: Jan 14, 2013
  - Compressed discordant read extraction + read renaming into one stage. This
    saves time spent on compressing/decompressing reads, but increases CPU
    load.

1.0.4: Jan 22, 2013
  - Replaced sorting of bowtie2 output during prograssive mapping.

1.1: Jan 24, 2013
  - Changed internal read naming scheme in order to speed up merging. Old fastq
    and sam files generated by alu-detect must be converted to the new format
    when used with the new version.

1.2: Feb 27, 2013
  - Reads to remap are now split into files by read group. This increases
    compatibility with tegetype. Added option SINGLE_READS_TO_REMAP to use
    existing files.

1.3: May 05, 2013
  - Introduced new wrapper script alu-detect to controll access to
    adding a reference, running detect, filtering. Various internal fixes.